GIST Scholar
Not a member yet
30271 research outputs found
Sort by
Bias-Free Solar Upcycling of Nitrate and Glycerol with Highly Efficient and Durable Organic Semiconductor-Based Photoelectrodes
No full textThe sustainable management of biodiesel byproducts, nitrate (NO3-) and glycerol, remains a critical environmental challenge, disrupting global nitrogen and carbon cycles. Addressing this issue from a materials science perspective, a bias-free photoelectrochemical (PEC) upcycling system based on the integration of functional energy materials is presented. A nickel-iron-phosphorus (Ni-Fe-P) electrocatalyst exhibits synergistic bi-functional activity for selective NO3- reduction and glycerol oxidation, driven by the redox dynamics of Ni and Fe and electronic modulation by phosphorus incorporation. Through a metal-foil encapsulation strategy, the catalyst is integrated with organic semiconductor (OS)-based photoanodes and photocathodes, forming a highly efficient OS-based PEC system capable of stable operation under continuous solar illumination. The individual photoelectrodes demonstrate photocurrent densities of +15.7 and-14.8 mA cm(-2) under AM 1.5G conditions, while exhibiting remarkable stability with over 96% of initial performance retained after 60 h. To enable bias-free solar upcycling, a dual-photoelectrode PEC configuration is constructed, delivering a high reaction current of 11.04 mA cm(-2) along with >95% Faradaic efficiencies and excellent selectivity for both ammonia (NH3) and formic acid (FA) production. This work underscores the potential of materials-driven PEC platforms for selective solar chemical upcycling.TRUEsciescopu
Coherence analysis of local randomness and nonlocal correlation through polarization-basis projections of entangled photon pairs
No full textPolarization-entangled photon pairs generated from second-order nonlinear optical media have been extensively studied for both fundamental research and potential applications of quantum information. In a spontaneous parametric down-conversion process, quantum entanglement between paired photons, often regarded as ‘mysterious,’ has been demonstrated for local randomness and nonlocal correlation through polarization-basis projections using linear optics (Phys. Rev. A 60, R773 (1999)). Unlike the conventional particle nature-based perspective, this paper presents a coherence analysis rooted in the wave nature of quantum mechanics to examine these well-established quantum phenomena, incorporating polarization control of the paired photons and their projection measurements. First, we analyze the quantum superposition of photon pairs generated randomly from cross-sandwiched nonlinear media, focusing on local randomness, which depends on the incoherence among measured events. Second, we investigate coincidence detection between paired photons to understand the nonlocal correlation arising from independently controlled remote parameters, resulting in an inseparable product-basis relationship. This coherence-based approach sheds light on a deterministic perspective on quantum features, emphasizing the significance of phase information intrinsic to the wave nature of photons, which is incompatible with the particle nature. © The Author(s) 2025.TRUEsciescopu
Anti-proteolytic regulation of KRAS by USP9X/NDRG3 in KRAS-driven cancer development
No full textCancers with activating mutations of KRAS show a high prevalence but remain intractable, requiring innovative strategies to overcome the poor targetability of KRAS. Here, we report that KRAS expression is post-translationally up-regulated through deubiquitination when the scaffolding function of NDRG3 (N-Myc downstream-regulated gene 3) promotes specific interaction between KRAS and a deubiquitinating enzyme, USP9X. In KRAS-mutant cancer cells KRAS protein expression, downstream signaling, and cell growth are highly dependent on NDRG3. In conditional KrasG12D knock-in mouse models of pancreatic ductal adenocarcinoma, Ndrg3 depletion abolishes Kras protein expression and suppresses intraepithelial neoplasia formation in pancreas. Mechanistically, KRAS protein binds to the C-terminal serine/threonine-rich region of NDRG3, subsequently going through deubiquitination by USP9X recruited to the complex. This interaction can be disrupted in a dominant-negative manner by a C-terminal NDRG3 fragment that binds KRAS but is defective in USP9X binding, highly suppressing KRAS protein expression and KRAS-driven cell growth. In summary, KRAS-driven cancer development critically depends on the deubiquitination of KRAS protein mediated by USP9X/NDRG3, and KRAS-addicted cancers could be effectively targeted by inhibiting the KRAS-NDRG3 interaction. © 2025. The Author(s).TRUEsciescopu
High-Frequency Flux Controlled MOS-Based Floating Meminductor Emulator Maryaradhiya Daimari School of Information and Computing Gwangju Institute of Science and Technology
No full textThis thesis presents a novel MOS-based meminductor emulator designed for high-frequency applications, with a focus on addressing fabrication challenges through emulation techniques. Utilizing 180 nm CMOS technology, the proposed emulator operates up to 10 MHz, with post-layout simulations demonstrating low area consumption (878.40 µm²) and power dissipation (0.47 mW). The emulator's architecture, which employs an Operational Transconductance Amplifier (OTA) and capacitors, successfully replicates the meminductor's pinched hysteresis loop in the current-flux plane, a key characteristic for memory behaviour. In addition, the study demonstrates the practical relevance of the proposed meminductor through its integration into a high-speed meminductor-based leaky integrate-and-fire (M-LIF) neuron. This neuron circuit, designed for neuromorphic pattern classification, achieves a spiking rate of 70 million spikes per second and an energy efficiency of 0.090 pJ/spike, outperforming existing designs in both speed and power. The system successfully implements digit classification using a 5×1 neuron array with minimal hardware overhead and no passive components. The emulator’s performance is also benchmarked against existing designs in terms of area, frequency, and power, highlighting significant improvements in scalability and efficiency. These results underscore the emulator’s suitability for energy-constrained applications, including adaptive filters, chaotic signal generators, and neuromorphic computing. Overall, this work advances the integration of meminductive behaviour into CMOS-compatible systems, paving the way for future monolithic IC designs in analog computing and intelligent communication platforms.MasterI. Chapter 1. INTRODUCTION 11
1. 1. Motivation 11
1. 2. Thesis Organization 12
II. Chapter 2. BASIC OF MEM-ELEMENTS 13
2. 1. Memristor 14
2. 2. Memcapacitor 15
2. 3. Meminductor 15
III. Chapter 3. PROPOSED MOS-BASED MEMINDUCTOR 17
3. 1. Proposed Circuit Design 17
3. 1. 1. Operational Transconductance Amplifier (Integral Block) 17
3. 1. 2. Functional Block 18
3. 2. Proposed Meminductor 19
3. 3. Proposed Meminductor Application in M-LIF Neuron 22
3. 3. 1. Pattern Detection based on M-LIF Neuron 26
IV. Chapter 4. ANALYSIS OF SIMULATION AND EXPERIMENTAL RESULTS 30
4. 1. Simulation Results 30
4. 1. 1. Parametric Analysis 31
4. 1. 2. Temperature Variation 32
4. 1. 3. Process Variation 33
4. 1. 4. Monte Carlo Simulation 34
4. 1. 4. Post-Layout Simulation 35
4. 2. Experimental Results 36
4. 3. Comparison with Prior Works 37
V. Chatpter 5. CONCLUSION 41
5. 1. Summary 41
5. 1. Future Work 41
REFERENCES 43
CURRICULUM VITAE 4
Comparison of subcooled flow boiling in concentric annuli with bare and finned surfaces
No full textIn this study, the hydrodynamic and thermal characteristics of subcooled flow boiling in concentric annuli with inner heating wires, featuring bare and finned surfaces designed for heat transfer enhancement, were experimentally investigated, with a focus on cooling technology for EV charging cables. The bare wire test module consisted of a 300 mm long wire, with a 290 mm long inner heating surface and a diameter of 10 mm. The outer tube was constructed from transparent polycarbonate with a 22 mm diameter, allowing optical access to observe the flow. The finned wire test module has an identical construction, except for the addition of six longitudinally extruded fins with cross-sectional dimensions of 2 × 5 mm2. Experiments and flow visualizations, aided by high speed imaging, were conducted using HFE-7100 as the working fluid at mass velocities, G, 463.7 to 1,309.6 kg/m2s and heat fluxes from 0.90 to 9.93 W/cm2, under inlet pressures between 120 and 240 kPa. The study investigated thermal non-equilibrium phenomena, driven by asymmetric bubble accumulation, and their effects on circumferential temperature distribution, which showed opposing trends between the bare and finned surfaces. An evaluation of previous empirical correlations for predicting subcooled flow boiling heat transfer coefficients was performed, followed by an assessment of fin performance in enhancing heat transfer. A comparative analysis with the bare surface module of the same dimensions confirmed that the longitudinal fins provided superior thermal management, maintaining the surface temperature below the safety threshold, Ts,safe. The longitudinal fins demonstrated a 54.5 % reduction in charging time and a 183.7 % increase in charging current for charging up to 80 % state of charge (SOC) of a 77.4 kWh battery, compared to the bare wire. © 2024 Elsevier LtdFALSEsciescopu
A Construction of Connected Dominating Set with Predetermined Nodes and Vision-based Formation control with Control Barrier Function
No full text이 논문은 분산 알고리즘에 대한 두 가지 주요 주제를 다룬다.
첫 번째 주제는 무선 애드혹 네트워크에서의 문제를 다루며, 특히 연결 지배 집합(CDS)을 가상 백본 네트워크로 구성하는 데 중점을 둔다. 기존 연구들이 모든 노드가 균일한 자격을 갖춘 것으로 가정한 반면, 본 논문은 고유한 특성으로 인해 CDS에 포함되거나 제외되어야 하는 특정 노드가 사전에 결정된 상황을 고려한다. 이를 바탕으로 강제-CDS 문제를 정의하고, 사전 결정된 노드를 포함하는 CDS 구성 문제로 확장하였다. 또한 기존의 2홉 접근 방식을 넘어, 1홉 정보를 활용한 중앙 집중식 및 분산식 알고리즘을 새롭게 제안하였다.
두 번째 주제는 로봇 간 통신에 의존하지 않고 가시성 제약 조건하에서 작동하는 비홀로노믹 이동 로봇을 위한 비전 기반 편대 제어 알고리즘이다. 선행 에이전트를 추적하는 데 사용되는 카메라가 시야각 제한을 가지므로, 효과적인 추적을 위해 실시간 가시성을 유지하는 것이 중요하다. 이를 해결하기 위해 카메라의 픽셀 좌표를 기반으로 한 배리어 함수를 설계하였고, 이 함수를 제약 조건으로 활용하는 이차 최적화 기반 제어기를 통해 가시성 유지와 비홀로노믹 특성을 동시에 만족시킬 수 있음을 입증하였다. 제안된 방법의 유효성은 시뮬레이션을 통해 가시성을 유지하면서 원하는 편대 형상을 달성하는 성능을 확인함으로써 검증하였다. ©2025 원승범 ALL RIGHTS RESERVED|This dissertation addresses two topics about distributed algorithms and is divided into two parts. The first part addresses the challenges in wireless ad-hoc networks, particularly focusing on the construction of a Connected Dominating Set (CDS) as a virtual backbone network.
Unlike previous works that assume uniform qualifications for all nodes, we consider cases with predetermined nodes that must be either included in or excluded from the CDS due to their unique functionalities. In this context, we extend the Enforced-CDS problem to a CDS construction problem with predetermined nodes. Contribution includes the development of both centralized and distributed algorithms that utilize one-hop information, diverging from the traditional two-hop approach. The size of the computed CDS by the proposed algorithms does not exceed 6.798 times the optimum.
The second part explores a vision-based formation control algorithm for nonholonomic mobile robots operating under visibility constraints without relying on inter-robot communication. The camera used to monitor the leading agent has field of view (FOV) constraints, making it crucial to maintain real-time visibility for effective leading agent tracking. To address these challenges, we designed a barrier function based on the camera's pixel coordinates and showed that the control barrier function based quadratic programming satisfies both the visibility maintenance and nonholonomic property. The effectiveness of the proposed method is validated through simulations that demonstrate its ability to maintain visibility and achieve the desired formation. ©2025 Seung-Beom Won ALL RIGHTS RESERVEDMasterAbstract (English) i
Abstract (Korean) ii
List of Contents iii
List of Figures v
I Introduction 1
1 Backbone Network for Wireless Ad-hoc Network 2
1.1 Research Background 2
1.2 Related Work 3
1.2.1 Centralized Algorithm 3
1.2.2 Distributed Algorithm 4
1.3 Motivation and Contribution 4
2 Vision-based Foramtion Control 6
2.1 Research Background 6
2.2 Related Work 6
2.3 Motivation and Contribution 7
II A Construction of Connected Dominating Set with Predetermined
Nodes 9
3 Theoretical Background 10
3.1 Terminologies and Notations 10
3.2 Definitions 11
4 CDS Construction Algorithm and Simualtion Analysis 14
4.1 Centralized-CDS Construction Algorithm 14
4.1.1 Algorithm Description 14
– iii –
4.1.2 Convergence Analysis 18
4.1.3 Performance Analysis 20
4.2 Distributed-CDS Construction Algorithm 21
4.2.1 Preliminaries 21
4.2.2 Algorithm Description 23
4.2.3 Performance Analysis 25
4.3 Simulation and Analysis 26
III Vision-based Formation Control with Control Barrier Function 29
5 Theoretical Background 30
5.1 Notations and Graph Representation 30
5.2 Acyclic Minimally Persistent Graph 30
5.3 System Modeling 32
5.3.1 Robot Model 32
5.3.2 Monocular Camera Model 33
5.4 Time Varying Control Barrier Function 35
6 Control Algorithm Design and Simulation Results 37
6.1 Visibility Constraint 37
6.2 Vision-based Formation Control Algorithm 38
6.3 Simulation Results 39
IV Conclusions 43
7 Summary and Future Work 44
7.1 A Construction of Connected Dominating Set with Predetermined Nodes . 44
7.2 Vision-based Formation Control with Control Barrier Function 44
Bibliography 4
Generalized Self-Play Reinforcement Learning for Othello under Dynamic Board Constraints
No full textCavity engineering of solid-state materials without external driving
No full textConfining electromagnetic fields inside an optical cavity can enhance the light-matter coupling between quantum materials embedded inside the cavity and the confined photon fields. When the interaction between the matter and the photon fields is strong enough, even the quantum vacuum field fluctuations of the photons confined in the cavity can alter the properties of the cavity-embedded solid-state materials at equilibrium and room temperature. This approach to engineering materials with light avoids fundamental issues of laser-induced transient matter states. To clearly differentiate this field from phenomena in driven systems, we call this emerging field cavity materials engineering. In this review, we first present theoretical frameworks, in particular, ab initio methods, for describing light-matter interactions in solid-state materials embedded inside a realistic optical cavity. Next, we overview a few experimental breakthroughs in this domain, detailing how the ground state properties of materials can be altered within such confined photonic environments. Moreover, we discuss state-of-the-art theoretical proposals for tailoring material properties within cavities. Finally, we outline the key challenges and promising avenues for future research in this exciting field. (c) 2025 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.FALSEsciescopu
Long-term evolution and synoptic meteorological modulation of PM2.5 and PM10 in Seoul
No full textAddressing particulate matter (PM) pollution in megacities like Seoul is crucial for public health and environmental sustainability, necessitating a comprehensive understanding of its long-term evolution and meteorological drivers. This study investigated the key factors affecting long-term PM concentrations in Seoul, South Korea, from 2000 to 2021, with a focus on the winter (DJF) and spring (MAM) seasons. To address the gap in PM research caused by the shorter observation period of PM2.5 compared with PM10, we used an extended PM2.5 dataset. This enabled a detailed analysis of PM2.5 and its relationship with PM10, which, despite some differences, generally displayed similar variability. Both PM2.5 and PM10 exhibited decreasing trends in winter as well as spring, although the rate of decline slowed in the last decade (2011-2020) compared with the earlier decade (2000-2010). Both seasons exhibited a strengthened interannual correlation between PM2.5 and PM10 in the last decade. Daily PM2.5 and PM10 levels generally fluctuated in a similar pattern in both seasons, which can be attributed to synoptic-scale meteorological systems, particularly migratory systems from Northwest China, which can remain stationary over Korea for several days, particularly in winter. This pattern continues into spring, albeit with a lower intensity. These findings provide valuable insights into PM2.5 variability and its correlation with PM10 over time, which may inform future PM2.5 mitigation strategies.FALSEsciescopu
Enhancing vision-language models through pre-training-free knowledge fusion with TransferCVLM
No full textRecent large vision-language multimodal models pre-trained with huge amount of image-text pairs show remarkable performances in downstream tasks. However, the multimodal pre-training has limitations in terms of resources and training time when it comes to obtaining new models that surpass existing models. To address these issues, we propose TransferCVLM, a method for efficiently constructing an advanced vision-language model without extensive multimodal pre-training. TransferCVLM integrates existing pre-trained unimodal models and a cross-modal fusion module into a combinative vision-language model (CVLM). For each task application, the CVLM is fine-tuned and further enhanced through knowledge distillation, where multimodal knowledge from a teacher vision-language model is transferred to the CVLM. We demonstrate that (1) the fine-tuned CVLM performs comparable to other vision-language models of similar size, that (2) the multimodal knowledge transfer consistently enhances the CVLM, and the knowledge-transferred CVLM outperforms the teacher multimodal model in most downstream tasks, and that (3) TransferCVLM can also be used for model compression when using small-size unimodal models, achieving better retainability than existing pre-training-based knowledge distillation methods. We estimate that the training of TransferCVLM takes only 6% of pre-training of other vision-language models. Our code is available at https://github.com/DMCB-GIST/TransferCVLM.FALSEsciescopu